METHOD FOR GROUND SHADOW REDUCTION IN SPORT VIDEO

摘要

A method for reducing shadow on a ground in a sport scene is provided to detect a shadow area in real time and perform shadow reduction with respect to the extracted shadow area, thereby providing a user with more improved viewing environment. It is determined whether a ground is present or not by using an RGB(Red, Green and Blue) value within each image frame of a sport scene(S100). An image frame where the ground is present is divided into n x n blocks to construct a ground block map(S200). By checking temporal block difference of the constructed ground block map, a shot boundary is detected(S300). In the detected short boundary, a long shot or non-long short is determined(S400). If the long shot, it is determined whether a shadow area is present within the image frame(S500). If so, a shadow area is extracted by using a YU value within the image frame(S600). The extracted shadow area is divided into first and second shadow areas according to intensity of a shadow component(S700). After that, a shadow component is reduced by controlling luminance according to strength of the divided first and second shadow areas(S800).